Fake Account Identification Using Machine Learning Approaches Integrated with Adaptive Particle Swarm Optimization

Abstract

 It is customary for humans, bots, and other automated systems to generate new user accounts by utilizing pilfered or otherwise deceitful personal information. They are employed in deceitful activities such as phishing and identity theft, as well as in spreading damaging rumors. An somebody with malevolent intent may generate a substantial number of counterfeit accounts, ranging from hundreds to thousands, with the aim of disseminating their harmful actions to as many authentic users as possible. Users can get a wealth of knowledge from social networking networks. Malicious individuals are readily encouraged to take use of this vast collection of social media information. These cybercriminals fabricate fictitious identities and disseminate meaningless stuff. An essential aspect of using social media networks is the process of discerning counterfeit profiles. This study presents a machine learning approach to detect fraudulent Instagram profiles. This strategy employed the attribute-selection technique, adaptive particle swarm optimization, and feature-elimination recursion. The results indicate that the suggested adaptive particle swarm optimization method surpasses RFE in terms of accuracy, recall, and F measure.